College Football Preseason Polls: Fun To Look at, but Keep Them out of the BCS

This article is not even going to try and defend the college football preseason polls because quite honestly I see them as indefensible. At the least I believe the preseason polls are a hindrance to an already failed system.

I am a huge proponent of having a college football playoff system for the FBS to find out who the true national champion is, partially relying on the human polls with their built in emotions and bias is just not the way to go. But including polls that start in the preseason is just a disaster.

Some may say that preseason polls cause no harm, that it is fun to have something to debate about and it gives the regular season polls a starting point. Well, the proof is in the history.

First, how accurate are preseason polls?

Let’s start by taking a look at the past five years and seeing what the facts show. For research data I compiled the preseason polls provided by seven individual polling services some of which the BCS uses in its formula and I “graded” each service’s preseason poll against their own end of season poll.

I came up with a scoring system that awarded points if a school in a poll was within three positions from where the school was in the preseason poll and where they ended up at the end of the season in that same poll. I had a separate scoring matrix to just count how many schools that were in a preseason top-25 poll ended up in that same service’s post- season poll no matter what the start or finish positions were.

Over the past five seasons, grading a combined total of 35 college football preseason polls for just bulk accuracy, how many schools from the preseason top-25 poll ended up on the same services postseason top-25 poll?

- The best bulk scoring a poll did was to have 19 of their preseason top-25 teams end up on their own postseason top-25 poll. Of the 35 total polls this happened only once, so this poll was 76 percent correct. That was the best.

- The worst bulk scoring a poll did was to have 10, which worked out to being 40 percent correct.

- The average bulk score of the 35 preseason polls was 14.9 or just below 60 percent correct.

The facts show that the overall accuracy of the preseason polls is pretty dismal; they are on average correct under 60 percent of the time. But let’s now take a look at the precision accuracy of these same polls, how good are they at ranking a team in their preseason poll and then having that same team ranked within three positions of their starting position by the end of the season.

- The best precision scoring a poll did was to have six teams finish the season within three positions of where they started in the preseason, and this happened five times.

- The worst precision scoring a poll did was to have zero.

- The average precision score of the 35 preseason polls was 3.8, or about 15 percent.

Where the facts show that preseason polls are under 60 percent correct overall on predicting if a school belongs in the top 25, these same polls truly fail at predicting where a school fits within the top-25 poll, only getting close 15 percent of the time.

By the way, this past season Auburn was a preseason No. 23 while Oregon was ranked No. 11 in the preseason.

The political ramifications of preseason polls

A poll should be an unbiased calculation based on the theory that if team A played team B on a neutral field, who would win? You do this with every team and every combination until a cream rises to the top and you have a ranking. Unfortunately, many pollsters are swayed by the latest media darling, have a geographic or emotional bias, or just do not check all the facts. 

A quandary that pollsters get themselves into is that they don’t start with a clean slate every week, re-ranking the schools properly, thereby more quickly filtering the bad teams out and the good teams in. I am a programmer by profession and we have a saying: “garbage in, garbage out,” meaning that your output is only going to be as good as the data you start with. 

Applying this to the polls, if your preseason poll is off and you don’t start with a clean slate each week, then you are just regurgitating the same bad data each week that you started with.

How many times have we seen a preseason highly ranked team lose week after week and still linger in the top 25 until Week 4 or 5 when they finally disappear? The opposite is also true of a team that is not in the preseason top-25: they turn out impressive wins even over ranked teams yet they don’t crack the top 25 until Week 4 or 5, at which point they have no chance of properly moving up into a higher echelon.

Because of this practice of not starting with a clean slate each week, preseason polls are detrimental to deriving at an accurate top 25 by the end of the season because they start with flawed and inaccurate information. Remember, history shows that the preseason polls are wrong over 40 percent of the time to begin with.

My suggestion: If the FBS is not going to move to a playoff system and thereby allow schools to prove who is best on the field, then the BCS needs to discount the use of polls that start with preseason data. I am not saying do away with preseason polls altogether—I am as big of a college football geek as there is and I enjoy checking out the preseason polls—but the BCS still needs fixing.